Methane Oxidation and Methylotroph Population Dynamics in Groundwater Mesocosms

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Woods Hole Open Access Server

Environmental Microbiology (2020) 00(00), 00–00 doi:10.1111/1462-2920.14929

Methane oxidation and methylotroph population dynamics in groundwater mesocosms

Olukayode Kuloyo,1,2 S. Emil Ruff,1,3 Aaron Cahill,4 and other methylotrophs were roughly similar across Liam Connors,5 Jackie K. Zorz,1 all samples, pointing at transfer of metabolites from Isabella Hrabe de Angelis,1,6 Michael Nightingale,1 the former to the latter. Two populations of Bernhard Mayer1 and Marc Strous 1* Gracilibacteria (Candidate Phyla Radiation) displayed 1Department of Geoscience, University of Calgary, successive blooms, potentially triggered by a period Calgary, Alberta, Canada. of methane famine. This study will guide interpreta- 2Shell International Exploration and Production Inc, tion of future field studies and provides increased Westhollow Technology Center, Houston, TX, USA. understanding of methylotroph ecophysiology. 3Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA, USA. Introduction 4The Lyell Centre, Heriot Watt University, Edinburgh, Oil and natural gas extraction from organic-rich shale for- United Kingdom. mations has transformed the global energy outlook 5Biomedical Sciences Department, Faculty of Medicine, (Malakoff, 2014). More than 100,000 of oil and gas wells University of Calgary, Calgary, Alberta, Canada. completed in the United States and Canada over the past 6Multiphase Chemistry Department, Max Planck Institute decade were horizontally drilled and hydraulically fractured for Chemistry, Mainz, Germany. (Kerr, 2010; McIntosh et al., 2019). In some of these wells, well bore integrity failure leads to the unintentional subsur- Summary face release of natural gas—also known as fugitive methane or stray gas (Vidic et al., 2013; Darrah et al., 2014). Extraction of natural gas from unconventional hydro- Such release may be followed by gas migration via carbon reservoirs by hydraulic fracturing raises con- multiphase fluid flow, through geological profiles, toward cerns about methane migration into groundwater. Microbial methane oxidation can be a significant groundwater and the water-unsaturated vadose zone, ulti- methane sink. Here, we inoculated replicated, sand- mately resulting in atmospheric emissions (Cahill et al., packed, continuous mesocosms with groundwater 2019). Methane, the main component of natural gas, has from a field methane release experiment. The meso- a global warming potential 86 times greater than CO2 over cosms experienced thirty-five weeks of dynamic 20 years, and 25 times greater over 100 years (Shindell methane, oxygen and nitrate concentrations. We et al., 2009; Frankenberg et al., 2011). determined concentrations and stable isotope signa- During migration, methane may be oxidized by tures of methane, carbon dioxide and nitrate and methanotrophic and methylotrophic Bacteria and Archaea monitored microbial community composition of inhabiting the groundwater. Methylotrophs are microor- suspended and attached biomass. Methane oxidation ganisms oxidizing compounds with a methyl (-CH3) was strictly dependent on oxygen availability and led group, such as methane and methanol. Methanotrophs to enrichment of 13C in residual methane. Nitrate did refers to the subgroup of methylotrophs capable of meth- not enhance methane oxidation under oxygen limita- ane oxidation. In freshwater and marine environments tion. Methylotrophs persisted for weeks in the microbial methane oxidation is known to be a critical absence of methane, making them a powerful marker methane sink that limits methane emissions (Le Mer and for active as well as past methane leaks. Thirty-nine Roger, 2001; Knittel et al., 2005). Methane oxidation may distinct populations of methylotrophic bacteria were proceed aerobically in the presence of oxygen, or anaer- observed. Methylotrophs mainly occurred attached to obically with nitrate, sulphate, and oxidized forms of iron sediment particles. Abundances of methanotrophs and manganese (Conrad, 1996; Hanson and Hanson, 1996; Boetius et al., 2000; Orphan et al., 2002; Ettwig

Received 30 May, 2019; revised; accepted 25 January, 2020. *For et al., 2010; Haroon et al., 2013; Ettwig et al., 2016; Cai correspondence. Tel. +1 403 220 6604; E-mail [email protected] et al., 2018). Aerobic oxidation of methane may lead to

© 2020 The Authors. Environmental Microbiology published by Society for Applied Microbiology and John Wiley & Sons Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 2 O. Kuloyo et al. increased turbidity resulting from microbial growth, oxy- Groundwater microbial communities harbor members of gen limitation, anoxic conditions (Cahill et al., 2017), and, the Candidate Phyla Radiation and other unknown bacteria in theory, production of sulfide by microbial sulfate reduc- that may affect the fitness of methylotrophs by antagonistic tion. Thus, while bioremediation may limit methane emis- ecological interactions (Brown et al., 2015; Anantharaman sions to the atmosphere, it may also reduce groundwater et al., 2016; Cross et al., 2019). In the present study, we quality (Révész et al., 2010; Osborn et al., 2011). Under address these issues using laboratory mesocosms inocu- oxygen-limiting conditions, aerobic methanotrophs may lated with groundwater from our previous field injection shift to partially anaerobic respiration using nitrate experiment (Cahill et al., 2017). (Hoefman et al., 2014; Kits et al., 2015; Heylen et al., Five sets of triplicated mesocosms were run for 2016). Methanotrophs may also leak out metabolites, such 35 weeks to investigate how different environmental aqui- as methanol or acetate, which are then further oxidized by fer conditions selected for specific methanotrophic and other methylotrophic bacteria, which may use nitrate as methylotrophic populations and affected methane biore- electron acceptor (Nercessian et al., 2005; Chistoserdova mediation outcomes. We also investigated the potential et al., 2009; Takeuchi, 2019). In a recent controlled natural for nitrate to serve as an alternate electron acceptor for gas injection field experiment, microbial methane oxidation methane oxidation, persistence of methanotrophs during was shown to be strictly dependent on oxygen (Cahill famine periods and the effect of methane bioremediation et al., 2017, 2018; Steelman et al., 2017; Forde et al., on stable isotope fingerprints of methane, nitrate and car- 2019). No anaerobic oxidation of methane was observed, bon dioxide. We used 16S rRNA gene amplicon and a lack of oxygen led to persistent (i.e. up to 700 days sequencing and cell counting to quantify the abundance post injection) presence of methane in the aquifer. of methanotrophic and associated bacteria in response to Even though methane is often detected in groundwater dynamic methane and oxygen regimes, of both attached with reducing redox conditions (Darling and Gooddy, and suspended biomass. Our study will assist interpreta- 2006; Gorody, 2012; Humez et al., 2016), the literature tion of future field studies on microbial methane bioreme- on methane oxidation in groundwater is limited compared diation in groundwater and provides new insights on the to marine and freshwater sediments and mostly reliant on ecophysiology of methylotrophic bacteria with regard to geochemical and isotopic analyses of groundwater gas oxygen, methane and nitrate concentrations. and water samples (Van Stempvoort et al., 2005; Cheung et al., 2010; Jackson et al., 2013; Humez et al., 2015; Humez et al., 2016). Methane of biogenic origin, which Results may have migrated into groundwater, or was produced in Mesocosms and inoculation situ by Archaea naturally present in the aquifer, can be distinguished from thermogenic methane, by its isotopic To explore methane bioremediation potential and micro- composition since biogenic methane is more isotopically bial community response to different stray gas leakage depleted in 13C(δ13C typically between -50‰ and scenarios, five sets of triplicated, continuous flow, labora- -110‰, relative to Vienna Pee Dee Belemnite, VPDB) tory mesocosm experiments were set up (Fig. 1). All than thermogenic methane (δ13C typically between -25‰ mesocosms, static sand columns perfused with a contin- and -55‰) (Whiticar, 1999). However, interpretations of uous flow of medium, were inoculated with groundwater isotope compositions are not always straightforward, obtained from a previous field methane release experi- because if methane is being oxidized by microbes, ment (Cahill et al., 2017). Seven groundwater samples remaining methane may become more enriched in δ13C, obtained from two wells located immediately downstream leading to a pseudo-thermogenic signature (Whiticar, of the methane release site, sampled at depths between 1999). This 13C enrichment in the remaining methane is 2 and 8 m and between 55 and 333 days after the start caused by a slight preference of methanotrophs for the of methane release, were mixed and used to inoculate all lighter (12C) isotope. fifteen mesocosms simultaneously to ensure a homoge- Field studies have a number of other limitations. Most or neous distribution of microbial diversity. all samples for microbial analysis come in the form of water The microbial community in the groundwater, in the from wells, whereas a large part of the groundwater bacte- water flowing out of the mesocosms and attached to ria may be attached to particles or sediments within the mesocosm sediment particles, was profiled with 16S subsurface and therefore remain invisible. Furthermore, rRNA gene amplicon sequencing. Across all 234 samples, groundwater flow and gas migration in the subsurface are 4,486 unique amplicon sequence variants (ASVs) were not homogeneous and therefore, environmental conditions recovered (Supplementary data S1). Thirty-nine ASVs are partially unknown, adding uncertainty to statistical were affiliated with ten known methylotrophic genera and inferences about relationships between environmental were observed in at least three samples. Representatives conditions and the occurrence of bacteria of interest. of ten different methylotrophic genera, affiliated with

Fig. 1 Schematic representation of the mesocosm experimental setup.

Alpha-, Beta- and Gammaproteobacteria, were detected most abundant in the mesocosms directly after inocula- (Table 1), including five methanotrophic genera. Because tion (up to 11%). the amplicon sequences were quite short (~400 nucleo- This shift might be explained by differences in resilience tides), classification beyond the level of genus was gener- across populations during 9-14 months of storage of the ally not feasible. Forty-three ASVs were affiliated with the groundwater between sampling and inoculation or by dif- Candidate Phyla Radiation. Two of those, both affiliated ferences in the ability to attach to sand particles during the with Gracilibacteria (BD1-5/SNO2), were occasionally quite inoculation itself. Methylobacter did persist in the meso- abundant (Table 1). Table 1 also lists two genera, repre- cosms and was the most abundant methanotroph in some sented by 51 ASVs, that were used as marker taxa for samples. Methylotrophic Alphaproteobacteria such as anoxic conditions: Pelosinus and Desulfosporosinus Methylocystis are considered more resilient during harsh (Beller et al., 2013; Shelobolina et al., 2007). Known conditions than methylotrophic Gammaproteobacteria anaerobic methane oxidizers were only detected sporadi- such as Methylobacter (Ho et al., 2012; Knief, 2015). cally, in very few samples and at extremely low abun- The first set of triplicated mesocosms was perfused dance. Therefore, it appears that methane oxidation in the with medium containing dissolved methane (up to mesocosms was strictly aerobic, as was observed previ- 0.4 mM) and oxygen (up to 0.15 mM). In the second set ously in the field (Cahill et al., 2017). of mesocosms, the medium contained nitrate (up to 0.3 mM), in addition to dissolved methane and oxygen. Amplicon sequencing showed that after inoculation, The third set was perfused with anoxic medium with dis- the microbial communities present in each mesocosm solved methane and nitrate. The fourth set received were similar to each other, but different from the ground- anoxic media with dissolved methane only. The final set water in the field. For example, whereas bacteria related received medium without any of these additions. to the Gammaproteobacterium Methylobacter were the Profiling the microbial community of cells suspended in most abundant methanotrophs in situ (up to 41 % relative the medium flowing out of the mesocosms was performed sequence abundance), bacteria affiliated with the at twelve time points during the 35 week experiments. Pro- Alphaproteobacteria Methylocystis or Methylosinus were filing the microbial community attached to sediments was

Table 1. Taxonomic afﬁliation and statistics of key amplicon sequence variants (ASVs). Supplementary Data S1 lists taxonomic classiﬁcations, sequences and abundances of all ASVs across all samples. niomna Microbiology Environmental # ASVs Field average Mesocosm average Mesocosm maximum Physiology Genus Class obser-ved abun-dance (%) abundance (%) abundance (%) #observations ASVs

Methanotroph Methylocystis/ Alpha- 3 0.0 2.4 32.3 196 15, 23, 82 sinusa proteobacteria Methanotroph Methylobacter Gamma- 5 14.5 0.1 6.1 56 55, 124, 272, 444, 1051 proteobaceria Methanotroph Methylovulum Gamma- 6 0.4 0.4 19.8 153 34, 40, 167, 402, 592, proteobaceria 1057 Methanotroph Methylomonas Gamma- 2 0.0 0.1 24.0 19 89, 187 proteobaceria ulse ySceyfrApidMcoilg n onWly&Sn Ltd., Sons & Wiley John and Microbiology Applied for Society by published Methanotroph Crenothrix Gamma- 2 0.0 0.0 0.8 11 437, 823 proteobaceria Methylotroph Hyphomicro-bium Alpha- 8 0.0 0.2 4.6 113 88, 228, 424, 853, 1115, proteobacteria 1227, 1831, 3223 Methylotroph Methylo-bacterium Alpha- 3 0.0 0.0 1.0 24 533, 710, 875 proteobacteria Methylotroph Methylo-versatilis Beta- 1 0.0 2.1 29.4 194 14 proteobacteria Methylotroph Methylotenera Beta- 6 2.1 0.2 3.6 122 106, 190, 327, 539, 646, proteobacteria 945 Methylotroph Methylophilus Beta- 3 3.0 0.1 13.5 14 111, 153, 1552 proteobacteria Unknown Gracilibacteria Candidate Phyla 2 0.0 0.6 12.1 150 24, 74 Radiation Fermentation Pelosinus Negativicutes 11 0.0 2.3 86.4 94 10, 19, 707, 1452, 1592, 2442, 2550, … Sulfate Desulfosporosinus Clostridia 40 2.4 0.3 19.8 162 46, 351, 374, 381, 382, reduction 457, 525, 724, … niomna Microbiology Environmental a. These two genera could not be discriminated based on the 400 nucleotide amplicon. Methane oxidation in groundwater mesocosms 5 much more disruptive, because mesocosms needed to be carbon dioxide and nitrate were internally consistent opened and sediment removed. This was only done three across all samples. For example, samples with lower times throughout the experiment. methane concentrations showed higher enrichment of 13C in residual methane, as well as higher carbon dioxide concentrations and higher enrichment of 12C in produced Mesocosm biogeochemistry carbon dioxide. Samples with lower nitrate concentrations During the first ten weeks, the methane concentration in showed higher enrichment of 15N in the residual nitrate. the medium flowing into the mesocosms was 0.4 mM. The medium flow rate was 100 ml day-1, equivalent to a water -1 Methylotrophs attached to sediment particles velocity of 1.8 m day . During this “low CH4” period, those mesocosms supplied with dissolved air generally dis- The outflowing medium of all mesocosms contained 5 -1 played complete methane consumption (Fig. 2A). They 1.3Æ0.3Á10 cells ml (SD, n = 30) in the “low CH4” also displayed residual oxygen in the outflowing medium phase of all experiments (Supplementary Data S1). Cell (Fig. 2E), indicating that those mesocosms were generally counts of suspended cells peaked at 5.1Á105 and 3.8Á105 -1 oxic. In mesocosms without dissolved air, the methane cells ml during the “high CH4” phase of mesocosms with concentrations in the outflowing medium were higher, but and without dissolved air respectively. An additional still lower than in the inflowing medium, indicating some unknown number of cells inhabited the mesocosms methane consumption in these experiments. Given the attached to sediment particles. Attached cells were sam- absence of known anaerobic methanotrophs in our pled for community profiling during week 10, at the end of amplicon data set, methane consumption in these meso- the “low CH4” phase and in weeks 31 and 35, at the end cosms can most easily be explained by assuming some of the “high CH4” phase and during the last methane- oxygen ingress, for example via rubber tubing (Fig. 1), into famine phase. The sediment communities displayed these mesocosms. Nitrate was not measured during the more diversity than the suspended communities, were

“low CH4” period (Fig. 2C). more stable in time and more similar across replicates After ten weeks, methane was removed from the and treatments (Fig. 3). They also displayed two to medium of all mesocosms, for seven weeks, enabling us twenty times higher relative sequence abundance of met- to assess persistence of methylotrophic bacteria in the hanotrophs and other methylotrophs (Fig. 3B). All absence of methane. methyotrophic genera showed higher sequence abun- Seventeen weeks after the start of the experiment, dance in sediments than in water, except the methane (~1 mM) was reintroduced into the medium, and methanotrophic Gammaproteobacterium Methylovulum. the medium flow rate was gradually increased to 200 ml Populations affiliated with that genus displayed slightly day-1, equivalent to a water velocity of 3.6 m day-1. During higher average sequence abundance among suspended this 16-week “high CH4” period, methane was present in cells, but the difference was not significant. Two signa- excess (Fig. 2A), and oxygen limitation occurred (Fig. 2E). ture anaerobic genera, Pelosinus and Desulfosporosinus, Methane was detected in the outflowing medium at a con- were more abundant among suspended cells than in centration up to 0.4 mM. In the presence of oxygen, the attached communities (15×,5×, respectively). This indi- δ13C value of this residual methane was between -27 ‰ cated the biofilms attached to sand particles did not fea- and -12 ‰ (Fig. 2B), more than 10 ‰ higher than the δ13C ture steep oxygen gradients, because such gradients value of methane of -36 ‰ in the inflowing medium. At the would have led to higher abundances of these anaerobes same time, the δ13C value of the dissolved carbon dioxide in attached communities. The two Gracilibacteria ASVs decreased from -28‰ in the inflowing medium to below #24 and #74 displayed 1.5- and 6-times higher abun- -43 ‰ in the outflowing medium (Supplementary Data S1). dance in water samples, respectively. 13C enrichment in residual methane and 12C enrichment in dissolved carbon dioxide were much lower in mesocosms Role of methane, oxygen and nitrate in niche without oxygen, indicating that less methane was oxidized differentiation in those experiments. Mesocosms provided with nitrate displayed nitrate con- Differences in substrate ranges and affinity among taxa sumption, reproduced across replicates, especially in and strains are often invoked to explain differences in experiments that did not receive dissolved air (Fig. 2C). environmental abundances and are a key aspect of Consistently, the δ15N value of nitrate increased from canonical niche definitions of bacteria (Dunfield et al., 15 ‰ in the inflowing medium, up to 44 ‰ in the out- 1999; Dunfield and Conrad, 2000; Ho et al., 2012; flowing medium (Fig. 2D). The rate of nitrate consumption Hoefman et al., 2014; Knief, 2015). In our study, we varied, and it was not observed at all time points. Con- expected that methane, oxygen and nitrate availability centrations and isotopic compositions of methane, would be key drivers of community composition.

Fig. 2 Environmental conditions during the periods of low and high methane supply in four sets of triplicated mesocosms (the fifth set did not receive any methane, nitrate or air). The legend shows conditions and colors. A, The methane concentration in medium flowing out of the mesocosms. B, δ13C of the residual methane. C, Nitrate concentration in outflowing medium. D, δ15N of residual nitrate. E, Oxygen concentration in outflowing medium. Nitrate and isotope composition were only measured during the high methane phase. Each dot represents one sample, with values tabulated in Supplementary Data S1.

Surprisingly, sets of replicated mesocosms did not dis- with both dissolved air and nitrate, whereas Methylocystis play significant differences in overall richness (Fig. 4A) abundances were highest in mesocosms with dissolved nor composition (Fig. 4B). Community richness across all air only. Among other methylotrophs, Methylotenera mesocosms was significantly lower than in the field. sequence abundances were relatively stable across condi- Mesocosms supplied with dissolved air displayed slightly tions. Although individual ASVs associated with a single higher diversity than those without. Mesocosms supplied methylophilic genus showed slightly different trends, no with nitrate displayed slightly lower diversity than those consistent patterns were observed that pointed to niche without, but both differences were not significant. Non- differentiation among variants within genera. metric multidimensional scaling showed communities The sequence abundance of the anaerobic signature overlapping with each other, indicating that similar com- genera, Pelosinus and Desulfosporosinus were nega- munities were enriched in all mesocosms, independent of tively affected by both air and nitrate. These bacteria environmental conditions. were most abundant in the data sets of mesocosms with- Oxygen availability increased the relative sequence out methane, nitrate and air. abundance of methanotrophs, but not of other meth- The two Gracilibacteria ASV abundances displayed ylotrophs (Fig. 4C). Availability of nitrate did not signifi- different responses to environmental conditions. #24 cantly change the overall sequence abundances of was present at much higher sequence abundance methanotrophs and other methylotrophs (Fig. 4D). in mesocosms with dissolved air. #74 had almost Methylobacter, Methyloversatilis and Hyphomicrobium even sequence abundances across all five sets of sequence abundances where much higher in mesocosms mesocosms.

Fig. 3 | Comparison of sediment and water communities. A, Nonmetric multidimensional scaling (NMDS) plot showing the sediment community (red) was more stable across experiments and time than the water community (light blue). Size of bubbles shows Shannon entropy. Richness of the sediment communities (135Æ68 ASVs, Shannon 3.1Æ0.8) was higher than of the water communities (102Æ31 ASVs, Shannon 2.6Æ0.7), but lower than in the field (175Æ82 ASVs, Shannon 3.6Æ0.9). B, Relative sequence abundance of methanotrophs (18 ASVs detected) and all other methylotrophs (21 ASVs detected) was higher in sediment communities than water communities (Wilcoxon rank sum test). P values of significant differences between water and sediment abundances are shown. All methylotrophic bacteria showed higher sequence abundance in sediment communities, except for ASVs affiliated with Methylovulum.

Community turnover and succession relatively large amount of suspended cells (Fig. 3C), it might have been partially “washed out” of the meso- After transplantation of a natural community into a labora- cosms at the higher medium flow rate during the “high tory environment, adaptation and/or acclimat(izat)ion will CH ” phase. Among other methylotrophs, occur, and these processes may affect community func- 4 Hyphomicrobium displayed the highest increase in rel- tions such as methane oxidation (Poursat et al., 2019). ative sequence abundance during the “high CH ” This can also lead to turnover and succession of individ- 4 phase, while Methylotenera populations collapsed. ual populations. Over time, the mesocosms experienced Both signature anaerobic genera displayed much a loss of richness (Fig. 5A). During the 35 week experi- higher sequence abundances during the “high CH ” ments, about 50 % of the observed ASVs were lost. Time 4 phase, consistent with the onset of oxygen limitation displayed much stronger control over community struc- (Fig. 2E). Methyloversatilis was unique among all ture than differences in environmental conditions methylotrophs in that it maintained a high abundance between sets of mesocosms (Fig. 5B), as shown by in the amplicon data sets of mesocosms without clearly separated clusters for each phase of the experi- methane. ments in the nonmetric multidimensional scaling plot Both Gracilibacteria ASVs displayed strong temporal (Anosim R 0.68, significance 0.001). Because this turn- dynamics (Fig. 5D). ASV #74 was abundant during the over also occurred in mesocosms without methane, oxy- “low CH ” phase, bloomed during the first famine phase, gen or nitrate, it appeared to be unrelated to differences 4 and then its population collapsed for the remainder of the in methane, oxygen and nitrate availability experienced experiment. ASV #74 was succeeded by #24. ASV #24 by the other mesocosms. was undetectable at first, but bloomed during the “high Weeks of famine did not lead to significant changes CH ” phase. Its population collapsed during the second in relative sequence abundances for methanotrophs 4 famine phase. and other methylotrophs (Fig. 5C). However, methanotroph sequence abundance increased significantly during the “high CH ” phase. Among methanotrophs, 4 Discussion Methylobacter displayed the highest increase in relative sequence abundance. Methylovulum displayed a The U.S. Government recommends 10 mg L-1 (0.6 mM) slight decrease in relative sequence abundance during dissolved methane as the safety threshold value, above this phase. Because the latter population included a which action must be taken due to the risks of explosion

Fig. 4 Differences between microbial communities incubated with or without methane, oxygen and nitrate. A, Sediment community richness in the incubations was lower than in the field, but differences between conditions were not significant (Wilcoxon rank sum test). B, Nonmetric multidimensional scaling (NMDS) plot showing mesocosm communities were different from field communities but similar to each other across all experiments, independent of conditions (colors are the same as in panel a). C, Relative sequence abundance of methanotrophs was higher in experiments with oxygen (Kruskal-Wallis rank sum test). Sequence abundances of other methylotrophs were not significantly different. D, No significant differences in sequence abundances of methanotrophs and other methylotrophs were observed between experiments with and without nitrate.

Fig. 5 Changes of microbial community structure with incubation time. A, Sediment community richness decreased during the incubations. B, Nonmetric multidimensional scaling plot showing sediment communities displayed similar turnover across all experiments, independent of oxygen, methane or nitrate availability (Anosim R 0.68, significance 0.001). Note that panels A and B include the control, which received no methane, oxygen or nitrate. C, Relative sequence abundance of methanotrophs increased during high methane supply. The absence of methane (famine) periods did not lead to significant decline of methanotrophs. Changes in the abundance of other methylotrophs were not significant. D, Succession of two different Gracilibacteria (Candidate Phyla Radiation) populations, observed in the water (suspended cell) community. Signif- icant differences (Kruskal-Wallis rank sum test) between sequential time periods in C and D shown by solid black arrows with P values. Each dot represents one sample.

© 2020 The Authors. Environmental Microbiology published by Society for Applied Microbiology and John Wiley & Sons Ltd., Environmental Microbiology 10 O. Kuloyo et al. involved with out-gassing of methane and its accumula- large part of the methane was transferred from the met- tion (Eltschlager et al., 2001; Humez et al., 2016). During hanotrophs to the other methylotrophs in the form of the first ten weeks of our incubations, the mesocosms metabolites such as methanol and acetate, and/or that received medium containing 0.2‑0.4 mM dissolved meth- the other methylotrophs were more versatile, feeding on ane. Under laboratory conditions, this mimicked a minor substrates from other sources. The persistence of shallow aquifer methane contamination event. During this Methyloversatilis in experiments without methane was time, the measured dissolved oxygen concentration consistent with the latter possibility (Kalyuzhnaya et al., (0.01-0.13 mM) was generally sufficient to realize com- 2006). In any case, the overall methylotroph abundance plete methane oxidation. was not significantly stimulated by nitrate and nitrate did Between weeks 19 and 33, the mesocosms were sup- not enhance methane bioremediation. Rates of microbial plied with up to 1 mM or 16 mg L-1dissolved methane, nitrate reduction were observed to be somewhat erratic mimicking a contamination above the safety threshold. and non-reproducible in our mesocosms, despite all other This led to incomplete, aerobic oxidation of methane. In conditions being well constrained and controlled. our previous field study (Cahill et al., 2017), anaerobic Methylobacter, Methyloversatilis and Hyphomicrobium methane oxidation did not appear to play a role in meth- might have consumed nitrate, because they appeared to ane bioremediation, and the same was true in the present benefit the most from nitrate addition. study. Because subsurface methane release displaces The replicated, controlled experimental design provided oxygen, and anaerobic methane oxidation can apparently some support for niche differentiation among met- not be taken for granted, methane bioremediation can hanotrophs and other methylotrophs. Methylobacter was become a slow process, taking hundreds of days, as was found to benefit from both oxygen limitation and nitrate previously shown (Cahill et al., 2017). Future studies may (SIMPER p < 0.05). This bacterium was the most abundant show to what extent it is possible to establish anaerobic in the field methane release experiment, which was charac- methane oxidation by bioaugmentation approaches terized by oxygen-limiting conditions. Among other meth- (Takuechi et al., 2004; Nikolopoulou et al., 2013; Dai ylotrophs, Methyloversatilis and Hyphomicrobium appeared et al., 2015). to benefit most from nitrate, even though Methylotenera Even in the absence of anaerobic methanotrophs, it was could in theory also compete for nitrate (Mustakhimov et al., still surprising that nitrate (supplied at 0.3 mM, 20 mg L-1) 2013). Methyloversatilis was most successful in the did not significantly improve methane bioremediation. absence of methane. All these observations are consistent Nitrate occurs naturally in groundwater typically at concen- with the current understanding of methylotroph physiology trations <10 mg L-1, with higher concentrations attributed (Kalyuzhnaya et al., 2006; Ho et al., 2012; Knief, 2015). to anthropogenic contamination from synthetic fertilizers Thus, amplicon sequencing could enable inferences about or manure in agricultural runoff (Rouse et al., 1999; environmental conditions based on known ecophysiological Wassenaar et al., 2006; Canadian Council of Ministers of niches of detected taxa. the Environment, 2012; Sebilo et al., 2013) or wastewater Ecological selection by methane, oxygen and nitrate sources. Its maximum allowable concentration (MAC) for concentrations occurred against a background of com- drinking water according to the Canadian Water Quality munity turnover that exerted much stronger pressure on Guidelines (CWQG) for the protection of aquatic life is overall community composition and was independent of - -1 45 mg NO3 L (Canadian Council of Ministers of the Envi- environmental conditions. This might explain the variabil- ronment, 2012). Aerobic methanotrophs have been ity in methylotroph abundances between replicates and observed to shift to a partially anaerobic metabolism, with shows that ecological interactions, including antagonistic nitrate replacing oxygen as terminal electron acceptor for interactions between microbial populations and viral pre- the later steps of the methanotrophic pathway (Hoefman dation, might be more important factors for ecological et al., 2014; Kits et al., 2015; Heylen et al., 2016). Aerobic success than consistent differences in ecophysiological methanotrophs are also known to hand off metabolites to niches. other methylotrophs (Nercessian et al., 2005; Chi- The ecological success of a population affiliated with stoserdova et al., 2009; Takeuchi, 2019). These then Gracilibacteria is the clearest example of the importance assimilate and/or further oxidize these metabolites with of ecological interactions in our study. These bacteria nitrate as electron acceptor (Mustakhimov et al., 2013). also bloomed during our previous field methane release Methanotrophs and their methylotrophic associates could experiment, after methane injection was stopped (Cahill also compete for oxygen under aerobic or oxygen-limiting et al., 2017). Thus, in both field and mesocosm studies, a conditions. period of famine appeared to stimulate growth of these Across all experiments and samples, the relative bacteria. Gracilibacteria is part of the Candidate Phyla sequence abundance of methanotrophs and other meth- Radiation (CPR) (Brown et al., 2015; Hug et al., 2016), ylotrophs were quite similar. This indicated that either a comprised uncultured bacteria, ubiquitous in

© 2020 The Authors. Environmental Microbiology published by Society for Applied Microbiology and John Wiley & Sons Ltd., Environmental Microbiology Methane oxidation in groundwater mesocosms 11 groundwater, and having “incomplete” central metabolism attached cells, with methanotrophs generally more abun- (Wrighton et al., 2012; Hanke et al., 2014; Dudek et al., dant in attached biomass. Methane oxidation resulted in 2017). Although their metabolic repertoire is very sparse, enrichment of 13C in residual methane, making this a the genomes of some Gracilibacteria have been shown strong signature for successful bioremediation, applicable to encode several proteins of the serine pathway for to field studies. Methylotrophic populations were shown formaldehyde assimilation and amino acid biosynthesis to persist in the absence of methane for many weeks. (Hanke et al., 2014) including serine hydro- This means that detection of such bacteria in groundwa- xymethyltransferase, formate-tetrahydrofolate ligase, and ter could point to active as well as a past methane con- bifunctional protein FolD. Among CPR Bacteria, tamination event. This study will guide interpretation of Gracilibacteria have been shown to use a different future field studies on microbial methane bioremediation genetic code and are believed to survive through a highly in groundwater and provides increased understanding on syntrophic or parasitic lifestyle (Hanke et al., 2014; Ana- the ecophysiology of methylotrophic bacteria with regard ntharaman et al., 2016; Probst et al., 2018). Cross et al. to varying oxygen, methane and nitrate concentrations. (2019) showed that related SN01 bacteria may be isolated by immuno-targeting cell surface proteins used by SN01 for attachment to prey bacteria. The blooming of Experimental procedures these bacteria during famine periods indicates they may at least benefit from the demise of other populations. Mesocosms. Five sets of triplicated mesocosms were Our study shows that a lack of detection of met- used for experiments. Each mesocosm consisted of a hanotrophs is a clear indicator for the absence of signifi- sand-packed, fused quartz column (inner diameter cant methane oxidation. However, because of their 15 mm, outer diameter 18 mm, height 400 mm, volume persistence, the presence of these organisms does not 71 ml). It was closed with a butyl rubber stopper at the necessarily indicate that methane is currently being oxi- top and bottom, which was fitted with a needle for supply dized. Instead, their presence might also result from per- and removal of medium (Fig. 1). The sand (premium play sistence after past methane-release events. Our results sand, The QUIKRETE Companies, Atlanta, GA, USA) further indicate that obtaining sediment cores would pro- was screened with a mesh No.40 to retain sand with par- vide a much more realistic assessment of methylotroph ticle size between 0.42 mm and 1 mm. Afterwards, the abundances than relying solely on water samples from sand was washed five times with ample sterile deionized groundwater wells. water, sterilized by autoclaving at 121C for 20 min and Stable isotopes compositions of methane, carbon diox- dried in an oven at 105C. The mesocosm experiments ide and nitrate proved extremely useful to demonstrate were performed over 35 weeks (Fig. 1). biological conversions and are highly recommended as a Mesocosm inoculation. The inoculum was obtained complimentary tool in field studies, where mass balances between November 2015 and June 2016 from a well- will be less useful to infer biological consumption. The characterized shallow freshwater aquifer research facility methane used in our study was of thermogenic origin located at the Canada Forces Base (CFB), Borden, with a δ13C value of -36 ‰. Following methane oxidation, Ontario, Canada (Cherry et al., 1983; Sudicky and Illman, the δ13C value of the residual methane was up to 10 ‰ 2011). The groundwater samples were collected during a higher, indicating an even stronger thermogenic methane controlled shallow groundwater methane release experi- isotope signature. Our results also clearly showed that ment (Cahill et al., 2017) at depths between 2 and 8 m occurrence of methanotrophy can mask biogenic isotope below a 1 m vadose zone from two monitoring wells M6 signatures of methane, as it results in the 13C enrichment and M7. These two wells were located 1 m apart in the of any remaining methane. This could lead to the incor- downstream groundwater flow direction from the methane rect inference of thermogenic origins for biogenic meth- injection point (Cahill et al., 2017). Sterile 1 L Nalgene ane in groundwater surveys (Whiticar, 1999). HDPE bottles were completely filled with groundwater In conclusion, our study showed that methane biore- and shipped in iced coolers to Calgary. Upon arrival mediation can be strictly dependent on oxygen availabil- (5-7 days after sampling), DNA extraction and Illumina ity, consistent with previous field work (Cahill et al., 16S rRNA gene amplicon sequencing was performed on 2017). Nitrate, at a low, environmentally meaningful con- 250 ml aliquots of each sample. The remainder was centration, and oxygen limitation favoured growth of bac- stored for 9-14 months at 4C until inoculation. Seven teria related to Methylobacter, the most abundant samples with a combined volume of 5 L were pooled into methanotroph in the field experiment. However, con- a sterile bottle and recirculated (0.5 Lday-1) over five trip- sumption of nitrate was episodic and overall, did not sig- licated mesocosms for 72 h. The direction of flow through nificantly stimulate methane oxidation. Large differences the mesocosms was from bottom to top, to displace air were observed between abundances of suspended and pockets out of the mesocosms during inoculation.

© 2020 The Authors. Environmental Microbiology published by Society for Applied Microbiology and John Wiley & Sons Ltd., Environmental Microbiology 12 O. Kuloyo et al. Incubation media. All mesocosms were continuously filtered through a 0.1 μmVCTPmembranefilter supplied with a mineral salts medium (Whittenbury et al., (MilliporeSigma, Billerica, MA, USA) attached to a sterile -1 1970), containing (g L ): MgSO4Á7H2O 1.0, CaCl2.6H2O 15 mL glass microanalysis filter holder (MilliporeSigma). 0.2, KH2PO4 0.27, Na2HPO4.2H2O0.72andNH4Cl 0.002. The remainder was used to determine microbial cell num- It also contained 0.5 mL L-1 of a solution containing (g L-1) bers (see below). For DNA extraction from mesocosm ferric ammonium citrate 1.0, sodium ethylene-diamine- sediments in weeks 10, 31 and 35, the mesocosms were tetra-acetate (EDTA) 2.0, 38% hydrochloric acid 3 mL L-1. opened at the top and approximately 1 g of sediment Finally, it contained 1 ml of (g L-1) sodium EDTA 0.5, (~ 20 mm) was removed from each mesocosm with a ster-

FeSO4.7H2O0.2,ZnSO4.7H2O0.01,MnCl2.4H2O0.003, ile spatula and transferred into a 15 mL falcon tube. All ﬁl- H3BO3 0.03, CoCl2.6H2O0.02,CaCl2.2H2O0.001, ters and sediments were stored at -20 C until DNA

NiCl2.6H2O 0.002 and Na2MoO4.2H2O 0.003. Nitrate was extraction. For measurements of dissolved gases, added as KNO3 as specified in the results section. This autoclaved, helium-flushed serum bottles (30 mL) were medium was prepared in magnetically stirred 10 L Schott first filled completely with the same media used for a given bottles with Teflon lids. Media vessels for the mesocosm mesocosm. This bottle was then placed between the experiments were sparged continuously, during the entire medium vessel and the mesocosm. It was fitted with a experiment, using mass flow controllers (Alicat Scientific, long 21 Gauge 0.8 × 50 mm needle (BD, Franklin Lakes, Tucson, AZ, USA), with a mixture of air, methane, and NJ, USA) and a short 25 Gauge 0.5 × 25 mm needle helium. Sparging started 24 h before medium was sup- (BD) such that inflowing media first passed through this plied to the mesocosms. The medium vessels were stirred serum bottle before it entered the mesocosm. A similar at 100 rpm. Gas was vented from the medium bottles via serum bottle (filled with deionized water instead of medium a water lock to prevent overpressure and backflow of air and chilled at 0C) was placed between the mesocosm into the bottles (Fig. 1). Triplicated mesocosms were sup- and the effluent vessel. The serum bottles were left to plied with sterile medium from the same feed bottle at equilibrate for at least 15 hours to ensure that approxi- a rate of 100 mL day-1 (~1.4 mesocosm volume changes mately five serum bottle volumes had flowed through. day-1, ~1.8 m day-1)duringthefirst seventeen weeks. Then, serum bottles were disconnected, and stored at 4C Between weeks 17 and 25, the rate was gradually until analysis, within 24 h after sample collection. increased to 200 mL day-1, maintained until the end of the Concentration and isotope ratio measurements. Dis- experiment. Media was pumped into the top of the meso- solved gas concentrations (CH4,C2H6,CO2,N2 and O2) cosm experiments using peristaltic pumps (Ismatec, were analyzed after separation from water using the Wertheim, Germany). Effluent (spent) medium from the static headspace equilibrium technique (Kampbell and mesocosms was collected in a 10 L Schott bottle. All Vandegrift, 1998) and measured on a Bruker 450 Natural mesocosms were maintained at room temperature (23C). Gas chromatograph with measurement uncertainties of All columns were covered with black polyethylene sheeting Æ 5%. Stable carbon isotope (δ13C, relative to VPDB, to prevent growth of phototrophs. Vienna PeeDee Belemnite) ratios of CH4 and CO2 were Measurement of dissolved oxygen concentrations and analyzed on a MAT 253 isotope ratio mass spectrometer pH. Dissolved oxygen concentrations were measured (IRMS) coupled to Trace GC Ultra and GC Isolink using oxygen sensor spots (OXSP5; Pyroscience, (Thermo Fischer Scientific, Waltham MA, USA), with an

Aachen, Germany) near the top and bottom of mesocosm error of <0.5‰ for CH4 and 0.3‰ for CO2. The nitrate columns. The sensor spots were connected to a concentration was measured as total oxidized nitrogen

FireSting O2 fiberoptic oxygen meter (Pyroscience) previ- using a Gallery Plus automated photometric analyzer ously calibrated to 100% and 0% oxygen according to (Thermo Fischer Scientific). The isotopic composition of manufacturer’s instructions. Media supplied to the meso- nitrate was determined on N2O generated by the denitri- cosms was also monitored at regular intervals for dis- fier technique (Casciotti et al., 2002, 2007), using a Delta solved oxygen with an inline flow-through cell fitted with V Plus IRMS coupled to a Finnigan MAT PreCon such a sensor spot. The sensor measures oxygen using (Thermo Fischer Scientific), with an accuracy of 0.3‰ 15 18 red light excitable materials that generate oxygen- and 0.7‰ for δ N-NO3 and δ O-NO3, respectively. dependent luminescence in the near infrared. The pH of Microbial cell numbers.Unfiltered effluent water sam- inflowing and outflowing medium was measured offline ples (0.5 mL) were mixed with 1.5 mL of sterile 1% phos- with a benchtop pH meter (Mettler Toledo, Columbus, phate buffered saline (PBS) and 108 μloffilter-sterilized OH, USA). 37% formaldehyde. The fixed samples were stored over- Sample collection from mesocosms.Effluent (spent) night at 4C and subsequently filtered through a sterile medium from all mesocosms was collected in 50 mL fal- 0.1 μm MilliporeSigma VCTP membrane filter. The filters con tubes covered with autoclaved aluminum foil, over- were washed twice with 1% PBS and dried with 1% night on ice. A 35 mL aliquot of the collected effluent was PBS/ethanol (1:1) solution. The filters were stored at

© 2020 The Authors. Environmental Microbiology published by Society for Applied Microbiology and John Wiley & Sons Ltd., Environmental Microbiology Methane oxidation in groundwater mesocosms 13 -20C until analysis. The cells were stained with DAPI Sequence Analyses. Sequenced libraries were ana- (4’, 6-diamidino-2-phenylindole) as described previously lyzed using dada2 following the DADA2 Pipeline Tutorial (Porter and Feig, 1980). The filters were viewed under a v1.12 (Callahan et al., 2016). Briefly, forward reads were Zeiss Axio Imager A.2 microscope (Carl Zeiss Micros- quality-trimmed to 275 bp and reverse reads to 215 bp. copy GmbH, Jena, Germany) equipped with a X-Cite Primer sequences (17 bp forward, 21 bp reverse) were 120 LED lamp (Lumen Dynamics, Mississauga ON, removed from the sequence reads. Reads with more than Canada) and Zeiss Axiocam 506 mono camera. Cell two expected errors were discarded (“maxEE=c(2,2)”). counts were averaged for 20 fields of view (Kepner Paired reads were merged with the default dada2::mer- et al., 1994). gePairs parameters. Chimeric sequences were removed DNA extraction. DNA from the 0.1 μm filters and sedi- using dada2::removeBimeraDenovo and species level ment samples was extracted with the FastDNA Spin Kit taxonomy was assigned using dada2::assignTaxonomy for Soil (MP BioMedicals, Santa Ana, CA, USA) and dada2::addSpecies with silva_nr_v132_train_set and according to manufacturer’s instructions. Extracted DNA silva_species_assignment_v132, which are based on the samples were quantified with a Qubit 2.0 Fluorometer Silva small subunit reference database SSURef v132 (Thermo Fischer Scientific) and stored at -20C until DNA (release date: Dec, 13 2017; Quast et al., 2013). The amplification. original dada2 output ASV-by-sample table was used to Illumina 16S rRNA gene sequencing.TheV3-V4 determine ASV richness and composition. Wilcoxon region of 16S rDNA was amplified in a single-step PCR signed rank-tests were performed with ggsignif,an using a KAPA HiFi HotStart reaction kit and primers extension to ggplot2. Shannon entropy was calculated Pro341F/Pro805R targeting prokaryotes (Takahashi from the ASV-by-sample table using subsampling, to et al., 2014) and including Illumina adapters (Illumina account for unequal sampling. Bray-Curtis dissimilarities Inc. San Diego, CA, USA). The primers, Pro341F (5´- (Bray and Curtis, 1957) between all samples were calcu- CCT ACG GGN BGC ASC AG-3´) and Pro805R (5´- lated and used for two-dimensional nonmetric multi- GAC TAC NVG GGT ATC TAA TCC-3´) complemented dimensional scaling (NMDS) ordinations with 20 random standard Illumina forward and reverse primers. PCR starts (Kruskal, 1964). All analyses were carried out with mixtures contained 0.1 μM of the forward primer, 0.1 μM the R statistical environment and the packages vegan of the reverse primer, 12.5 μlof2× KAPA HiFi HotStart (Dixon, 2003), ggplot2 (Wickham, 2009), as well as with Ready Mix (Kapa Biosystems, Wilmington, MA, USA) custom R scripts. and 1 μl of template DNA (~1 ng μL-1),madeupto25μl with nuclease-free water. PCR reaction conditions were as follows: initial denaturation at 95Cfor3min, Acknowledgements followed by 32 cycles of 95 C for 30 s, 55 Cfor45s, The authors acknowledge funding from the Alberta Inno- and 72 C for 60 s, followed by a final step of 72 Cfor vates Technology Futures (AITF), and University of Calgary 5 min. The amplicon products of triplicate PCR reactions Eyes High Doctoral Scholarships (O.O.K., J.K.Z.) and AITF/ were pooled and purified using 0.8× volume of AMPure Eyes High Postdoctoral Fellowships (S.E.R.), as well as the XP magnetic beads (Beckman Coulter, Indianapolis, IN, PROMOS Internship Abroad Scholarship by the German Academic Exchange Service (I.H.d.A.). Additional support USA) as per manufacturer’s instructions. DNA amplicon was provided by the Natural Sciences and Engineering fi libraries were prepared from puri ed PCR-products Research Council of Canada (NSERC), Strategic Project using an Illumina Nextera XT DNA Library Prep Kit Grant no. 463045-14, the Campus Alberta Innovation Chair (i5 adapters.) with Nextera XT index kit v2 (i7 adapters) Program (M.S.), Alberta Innovates, The Canadian Founda- as per the manufacturer’s instructions. Reaction condi- tion for Innovation (M.S.), the Alberta Small Equipment Grant tions of the second (index) PCR were as follows: 95C Program (M.S.) and an NSERC Discovery Grant for 3 min, followed by 10 cycles of 9C for 30 s, 5Cfor (M.S. and B.M.). 45 s, and 72C for 60 s, and a final step of 72Cfor 5 min. PCR products were purified with AMPure XP Data Availability Statement beads and quantified with a Qubit 2.0 Fluorometer. Amplicon libraries were normalized to 2 nM, pooled The authors declare that the data supporting the findings in equal volumes, denatured in 0.2 N NaOH and of this study are available within the article and its diluted with hybridization buffer according to the Nextera Supplementary Information. An online version of the sup- XT protocol. Paired-end sequencing (300 × 300 bp) plementary information is also available: (https://figshare. of libraries at 15 pM final concentration was com/articles/Figure_S1/8175473). DNA sequence data are performed on an Illumina Miseq instrument using manu- available in the NCBI database under accession number facturer’s reagents and according to the manufacturer’s PRJNA513134 (https://www.ncbi.nlm.nih.gov/bioproject/ instructions. PRJNA513134).

References groundwater wells in Alberta, Canada. Appl Geochem 25: 1307–1329. Anantharaman, K., Brown, C.T., Hug, L.A., Sharon, I., Chistoserdova, L., Kalyuzhnaya, M.G., and Lidstrom, M.E. Castelle, C.J., Probst, A.J., et al. (2016) Thousands of (2009) The Expanding World of Methylotrophic Metabo- microbial genomes shed light on interconnected biogeo- lism. Ann Rev Microbiol. 63: 477–499. chemical processes in an aquifer system. Nat Commun. Conrad, R. (1996) The global methane cycle: recent 7: 13219. advances in understanding the microbial processes Beller, H.R., Han, R., Karaoz, U., Lim, H., and Brodie, E.L. involved. Environ Microbiol Rep 1: 285–292. (2013) Genomic and physiological characterization of the Cross, K.L., Campbell, J.H., Balachandran, M., Campbell, A. chromate-reducing, aquifer-derived Firmicute Pelosinus G., Cooper, S.J., Griffen, A., et al. (2019) Targeted isola- – sp. strain HCF1. Appl Environ Microbiol 79:63 73. tion and cultivation of uncultivated bacteria by reverse Boetius, A., Ravenschlag, K., Schubert, C.J., Rickert, D., genomics. Nat Biotechnol 37: 1314–1321. Widdel, F., Gieseke, A., et al. (2000) A marine microbial Dai, Y., Ningning, L., Zhao, Q., and Xie, S. (2015) Bioreme- consortium apparently mediating anaerobic oxidation of diation using Novosphingobium strain DY4 for – methane. Nature 407: 623 626. 2,4-dichlorophenoxyacetic acid-contaminated soil d impact Bray, J.R., and Curtis, J.T. (1957) An ordination of upland on microbial community structure. Biodegradation. 26: forest communities of southern Wisconsin. Ecol. Mono- 161–170. – graphs 27: 325 349. Darling, W.G., and Gooddy, D.C. (2006) The hydro- Brown, C.T., Hug, L.A., Thomas, B.C., Sharon, I., geochemistry of methane: Evidence from English ground- Castelle, C.J., Singh, A., et al. (2015) Unusual biology waters. Chem. Geol. 229: 293–312. across a group comprising more than 15% of domain Bac- Darrah, T.H., Vengosh, A., Jackson, R.B., Warner, N.R., and teria. Nature 523: 208–211. Cahill, A.G., Steelman, C.M., Forde, O., Kuloyo, O., Ruff, S. Poreda, R.J. (2014) Noble gases identify the mechanisms E., Mayer, M., et al. (2017) Mobility and persistence of of fugitive gas contamination in drinking-water wells over- methane in groundwater in a controlled-release field lying the Marcellus and Barnett Shales. Proc Natl Acad experiment. Nat Geosci. 10: 289–294. Sci U S A 111: 14076–14081. Cahill, A.G., Parker, B.L., Mayer, B., Mayer, K.U., and Dixon, P. (2003) VEGAN, a package of R functions for com- Cherry, J. (2018) High resolution special and temporal reso- munity ecology. J Veg Sci 14(6): 927–930. Available from: lution of dissolved gasses during a controlled natural gas . https://doi.org/10.1111/j.1654-1103.2003.tb02228.x. release experiment. Sci Total Environ. 622-623: 1178–1192. Dudek, N.K., Sun, C.L., Burstein, D., Kantor, R.S., Aliaga Cahill, A.G., Beckie, R., Ladd, B., Sandl, E., Goetz, M., Goltsman, D.S., Bik, E.M., et al. (2017) Novel microbial Chao, J., et al. (2019) Advancing knowledge of gas migra- diversity and functional potential in the marine mammal tion and fugitive gas from northeast British Columbia, oral microbiome. Curr Biol 27: 3752–3762. Canada. Greenhouse Gas Sci Technol. 9: 134–151. Dunfield, P.F., and Conrad, R. (2000) Starvation alters the Cai, C., Leu, A.O., Xie, G.-J., Guo, J., Feng, Y., Zhao, J.-X., apparent half-saturation constant for methane in the type et al. (2018) Methanotrophic archaeon couples anaerobic II methanotroph Methylocystis strain LR1. Appl Environ oxidation of methane to Fe(III) reduction. ISME J. 12: Microbiol 66: 4136–4138. 1929–1939. Dunfield, P.F., Liesack, W., Henckel, T., Knowles, R., and Callahan, B.J., McMurdie, P.J., Rosen, M.J., Han, A.W., Conrad, R. (1999) High-affinity methane oxidation by a Johnson, A.J., and Holmes, S.P. (2016) DADA2: High- soil enrichment culture containing a type II methanotroph. resolution sample inference from Illumina amplicon data. Appl Environ Microbiol 65: 1009–1014. Nature Methods 13: 581–583. https://doi.org/10.1038/ Eltschlager, K.K., Hawkins, J.W., Ehler, W.C., and nmeth.3869. Baldassare, F.J. (2001) Technical Measures for the Inves- Canadian Council of Ministers of the Environment. (2012) tigation and Mitigation of Fugitive Methane Hazards in Canadian Water Quality Guidelines for the Protection of Areas of Coal Mining. U.S. Department of the Interior, Aquatic Life: Nitrate Ion. Canadian Council of Ministers of Office of Surface Mining p 129. the. Winnipeg, Manitoba: Environment. Ettwig, K.F., Butler, M.K., Le Paslier, D., Pelletier, E., Casciotti, K.L., Sigman, D.M., Hastings, M.G., Böhlke, J.K., Mangenot, S., Kuypers, M.M.M., et al. (2010) Nitrite-driven and Hilkert, A. (2002) Measurement of the oxygen isotopic anaerobic methane oxidation by oxygenic bacteria. Nature composition of nitrate in seawater and freshwater using 464: 543–548. the denitrifier method. Anal Chem. 74: 4905–4912. Ettwig, K.F., Zhu, B., Speth, D., Keltjens, J.T., Jetten, M.S. Casciotti, K.L., Böhlke, J.K., McIlvin, M.R., Mroczkowski, S. M., and Kartal, B. (2016) Achaea catalyze iron-dependent J., and Hannon, J.E. (2007) Oxygen isotopes in nitrite: anaerobic oxidation of methane. Proc Nat Acad Sci U S A analysis, calibration, and equilibration. Anal Chem 79(6): 113: 12792–12796. 2427–2436. https://doi.org/10.1021/AC061598H. Forde, O.N., Cahill, A.G., Mayer, K.U., Mayer, B., Simister, R. Cherry, J., Gillham, R., Anderson, E., and Johnson, P. L., Finke, N., et al. (2019) Hydro-biogeochemical impacts of (1983) Migration of contaminants in groundwater at a land- fugitive methane on a shallow unconfined aquifer. Sci Total fill: A case study: 2. Groundwater monitoring devices. Environ. 690:1342–1354. J Hydrol 63:31–49. Frankenberg, C., Aben, I., Bergamaschi, P., Dlugokencki, E. Cheung, K., Klassen, P., Mayer, B., Goodarzi, F., and J., Van Hees, R., Houweling, S., et al. (2011) Global Aravena, R. (2010) Major ion and isotope geochemistry of column-averaged methane mixing ratios from 2003 to fluids and gases from coalbed methane and shallow

2009 as derived from SCIAMACHY: Trends and variabil- environmental samples: past and present. Microbiol Rev ity. J Geophys Res 116: D04302. 58: 603–615. Gorody, A.W. (2012) Factors affecting the variability of stray Kerr, R.A. (2010) Natural gas from shale bursts onto the gas concentration and composition in groundwater. Envi- scene. Science 328: 1624–1626. ron Geosci 19:17–31. Kits, K.D., Klotz, M.G., and Stein, L.Y. (2015) Methane oxida- Hanke, A., Hamann, E., Sharma, R., Geelhoed, J.S., tion coupled to nitrate reduction under hypoxia by the Hargesheimer, T., Kraft, B., et al. (2014) Recoding of the Gammaproteobacterium Methylomonas denitrificans, stop codon UGA to glycine by a BD1-5/SN-2 bacterium sp. nov. type strain FJG1. Environ Microbiol 17: 3219–3232. and niche partitioning between Alpha- and Knief, C. (2015) Diversity and habitat preferences of culti- Gammaproteobacteria in a tidal sediment microbial com- vated and uncultivated aerobic methanotrophic bacteria munity naturally selected in a laboratory chemostat. Front evaluated based on pmoA as molecular marker. Front Microbiol 5: 231. Microbiol 6: 1346. Hanson, R.S., and Hanson, T.E. (1996) Methanotrophic bac- Knittel, K., Lösekann, T., Boetius, A., Kort, R., and teria. Microbiol Rev 60: 439–471. Amann, R. (2005) Diversity and distribution of Haroon, M.F., Hu, S., Shi, Y., Imelfort, M., Keller, J., methanotrophic archaea at cold seeps. Appl Environ Hugenholtz, P., et al. (2013) Anaerobic oxidation of meth- Microbiol 71: 467–479. ane coupled to nitrate reduction in a novel archaeal line- Kruskal, J.B. (1964) Nonmetric multidimensional scaling: a age. Nature 500: 567–570. numerical method. Mathematics 29: 115–129. Heylen, K., De Vos, P., and Vekeman, B. (2016) Draft Le Mer, J., and Roger, P. (2001) Production, oxidation, emis- genome sequences of eight obligate methane oxidizers sion and consumption of methane by soils: a review. Eur J occupying distinct niches based on their nitrogen metabo- Soil Biol 37:25–50. lism. Genome Announcements 4: e00421–e00416. Malakoff, D. (2014) The gas surge. Science 344: https://doi.org/10.1128/genomeA.00421-16 1464–1467. Ho, A., Kerckhof, F.M., Luke, C., Reim, A., Krause, S., McIntosh, J.C., Hendry, M.J., Ballentine, C., Haszeldine, R. Boon, N., and Bodelier, P.L.E. (2012) Conceptualizing S., Mayer, B., Etiope, G., et al. (2019) A critical review od functional traits and ecological characteristics of methane- state of the art and emerging approaches to identify oxidizing bacteria as life strategies. Environ Microbiol Rep fracking-derived gasses and associated contaminants in 5: 335–345. aquifers. Environ Sci Technol 53: 1063–1077. Hoefman, S., Van Der Ha, D., Boon, N., Vandamme, P., De Mustakhimov, I., Kalyuzhnaya, M.G., Lidstrom, M.E., and Vos, P., and Heylen, K. (2014) Niche differentiation in Chistoserdova, L. (2013) Insights into denitrification in nitrogen metabolism among methanotrophs within an methylotenera mobilis from denitrification pathway and operational taxonomic unit. BMC Microbiol. 14(83). methanol metabolism mutants. J Bac 195: 2207–2211. Hug, L.A., Baker, B.J., Anantharaman, K., Brown, C.T., Nercessian, O., Noyes, E., Kalyuzhnaya, M.G., Lidstrom, M. Probst, A.J., Castelle, C.J., et al. (2016) A new view of the E., and Chistoserdova, L. (2005) Bacterial populations tree of life. Nat Microbiol 1: 16048. active in metabolism of C1 compounds in the sediment of Humez, P., Mayer, B., Nightingale, M., Ing, J., Becker, V., Lake Washington, a Freshwater Lake. Appl Environ Jones, D., and Lam, V. (2015) An 8-year record of gas Microbiol 71: 6885–6899. geochemistry and isotopic composition of methane during Nikolopoulou, M., Eickenbusch, P., Pasadakis, N., baseline sampling at a groundwater observation well in Venieri, D., and Kalogerakis, N. (2013) Microcosm evalua- Alberta (Canada). Hydrogeol J 24: 109–112. tion or autochthonous bioaugmentation to combat marine Humez, P., Mayer, B., Nightingale, M., Ing, J., Becker, V., oil spills. N Biotechnol 30: 734–742. Kingston, A., et al. (2016) Occurrence and origin of meth- Orphan, V.J., House, C.H., Hinrichs, K.-U., McKeegan, K.D., ane in groundwater in Alberta (Canada): gas geochemical and DeLong, E.F. (2002) Multiple archaeal groups medi- and isotopic approaches. Sci Total Environ 541: ate methane oxidation in anoxic cold seep sediments. 1253–1268. Proc Natl Acad Sci U S A 99: 7663–7668. Jackson, R.E., Gorodi, A.W., Mayer, B., Roy, J.W., Ryan, M.C., Osborn, S.G., Vengosh, A., Warner, N.R., and Jackson, R.B. and Van Stempvoort, D.R. (2013) Groundwater protection (2011) Methane contamination of drinking water accompa- and unconventional gas extraction: The critical need for field- nying gas-well drilling and hydraulic fracturing. Proc Natl based hydrogeological research. Groundwater 51:488–510. Acad Sci U S A 108: 8172–8176. Kalyuzhnaya, M.G., De Marco, P., Bowerman, S., Porter, K.G., and Feig, Y.S. (1980) The use of DAPI for iden- Pacheco, C.C., Lara, J.C., Lidstrom, M.E., and tifying and counting aquatic microflora1. Limnol Oceanogr Chistoserdova, L. (2006) Methyloversatilis universalis gen. 25: 943–948. nov., sp. nov., a novel taxon within the Betaproteobacteria Poursat, B.A.J., Van Spanning, R.J.M., De Voogt, P., and represented by three methylotrophic isolates. Int J System Parsons, J.R. (2019) Implications of microbial adaptation Evol Microbiol 56: 2517–2522. for the assessment of environmental persistence of Kampbell, D.H., and Vandegrift, S.A. (1998) Analysis of dis- chemicals. Crit Rev Environ Ecol 49: 2220–2255. solved methane, ethane, and ethylene in ground water by Probst, A.J., Lad, B., Jarett, J.K., Geller-McGrath, D.E., a standard gas chromatographic technique. J Chromatogr Sieber, C.M.K., Emerson, J.B., et al. (2018) Sci 36: 253–256. Differential depth distribution of microbial function and Kepner, R.L., Pratt, J.R., and Pratt, J.R. (1994) Use of fluoro- putative symbionts through sediment-hosted aquifers in chromes for direct enumeration of total bacteria in the deep terrestrial subsurface. Nat Microbiol 3: 328–336.

Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., utilizable for a bioremediation of trichloroethylene-contami- Yarza, P., et al. (2013) The SILVA ribosomal RNA gene nation. Environ, Geol. 45: 891–898. database project: improved data processing and web- Van Stempvoort, D., Maathuis, H., Jaworski, E., Mayer, B., based tools. Nucl Acids Res 41: D590–D596. and Rich, K. (2005) Oxidation of fugitive methane in gro- Révész, K.M., Breen, K.J., Baldassare, A.J., and Burruss, R. und water linked to bacterial sulfate reduction. Groundwa- C. (2010) Carbon and hydrogen isotopic evidence for the ter 43: 187–199. origin of combustible gases in water-supply wells in north- Vidic, R.D., Brantley, S.L., Vandenbossche, J.M., central Pennsylvania. Appl Geochem 25: 1845–1859. Yoxtheimer, D., and Abad, J.D. (2013) Impact of shale Rouse, J.D., Bishop, C.A., and Struger, J. (1999) Nitrogen gas development on regional water quality. Science 340: pollution: an assessment of its threat to amphibian sur- 1235009. vival. Environ Health Perspect. 107: 799–803. Wassenaar, L.I., Hendry, M.J., and Harrington, N. (2006) Sebilo, M., Mayer, B., Nicolardot, B., Pinay, G., and Mariotti, A. Decadal geochemical and isotopic trends for nitrate in a (2013) Long-term fate of nitrate fertilizer in agricultural soils. transboundary aquifer and implications for agricultural Proc Natl Acad Sci U S A 110: 18185–18189. beneficial management practices. Environ Sci Technol 40: Shelobolina, E.S., Nevin, K.P., Blakeney-Hayward, J.D., 4626–4632. Johnsen, C.V., Plaia, T.W., Krader, P., et al. (2007) Geo- Whiticar, M.J. (1999) Carbon and hydrogen isotope system- bacter pickeringii sp. nov., Geobacter argillaceus sp. nov. atics of bacterial formation and oxidation of methane. and Pelosinus fermentans gen. nov., sp. nov., isolated Chem Geol Technol 161: 291–314. from subsurface kaolin lenses. Int J Syst Evol Microbiol Whittenbury, R., Phillips, K.C., and Wilkinson, J.F. (1970) 57: 126–135. Enrichment, isolation and some properties of methane- Shindell, D.T., Faluvegi, G., Koch, D.M., Schmidt, G.A., utilizing bacteria. J Gen Microbiol 61: 205–218. Unger, N., and Bauer, S.E. (2009) Improved attribution of Wickham, H. (2009) ggplot2: Elegant Graphics for Data climate forcing to emissions. Science 326: 716–718. Analysis. New York: Springer-Verlag. Steelman, C.M., Klazinga, D.R., Cahill, A.G., Endres, A.L., and Wrighton, K.C., Thomas, B.C., Sharon, I., Miller, C.S., Parker, B.L. (2017) Monitoring the evolution and migration of Castelle, C.J., VerBerkmoes, N., et al. (2012) Fermenta- a methane gas plume in an unconfined sandy aquifer using tion, hydrogen, and sulfur metabolism in multiple time-lapse GPR and ERT. J Contam Hydrol 205:12–24. uncultivated bacterial phyla. Science 337: 1661–1665. Sudicky, E.A., and Illman, W.A. (2011) Lessons Learned from a Suite of CFB Borden Experiments. Ground Water – 49: 630 648. Supporting Information Takahashi, S., Tomita, J., Nishioka, K., Hisada, T., and Nishijima, M. (2014) Development of a prokaryotic universal Additional Supporting Information may be found in the online primer for simultaneous analysis of Bacteria and Archaea version of this article at the publisher’s web-site: using next-generation sequencing. PLoS One 9: e105592. fi Takeuchi, M. (2019) Possible cross-feeding pathway of fac- Supplementary Data S1 (https:// g.com/articles/Figure_S1/ ultative methylotroph Methyloceanibacter caenitepidi 8175473). Concentrations of methane, oxygen, nitrate, car- fl fl Gela4 on methanotroph Methylocaldum marinum S8. bon dioxide in in owing and out owing media of all meso- PLoS One 14: e0213535. cosms, pH, and isotopic signatures of methane, carbon Takuechi, M., Nanba, K., Furuya, K., Nirei, H., and dioxide and nitrate, as well as relative sequence abundances Yoshida, M. (2004) Natural groundwater of a gas field for all amplicon sequence variants.